storwize-svc migration/retype may leave extra volume copies after cinder crash

[Avishay Traeger]

Cinder storage-assisted volume migration and retype work by creating a new copy of the volume, waiting for it to sync, and deleting the original copy. During the sync time (which may be long), if the Cinder process goes down for any reason, we will end up with a volume that has two copies, which wastes resources.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/73323

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/73523

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/73523
Committed: https://git.openstack.org/cgit/openstack/cinder/commit/?id=4cac1bd3227fd8b65744e75f0df018fb34bcb1c1
Submitter: Jenkins
Branch: master

commit 4cac1bd3227fd8b65744e75f0df018fb34bcb1c1
Author: LarryLiu <email address hidden>
Date: Fri Feb 14 15:46:07 2014 +0800

    Storwize/SVC: Change volume copy task to async

    If Cinder crashes during a migration or retype (where data is moved and
    the operation can take a long time), the storage ends up with multiple
    copies of the same volume which requires storage admin intervention.

    This patch maintain a list of pending operations which is backed up in
    admin metadata, and a periodic task reviews the list and removes volume
    copies whose copy operation completed. When Cinder comes up, check the
    admin metadata and rebuild the list.

    Change-Id: I6549712bb0083996faced89c2207a4c438ae953d
    Closes-Bug: #1278035